Thanks Marianne, and to all who replied with this.
I've set MEDIA_UNMOUNT_DELAY, confirmed that it is written to the NT
registry, restarted the appropriate services, and the problem still
persists. It takes about 2 seconds between one job finishing and another
starting, but the tape is still unloaded on completion of each backup set
(ie once per database). Is there anything else I should try?
On a related note, during my testing, I've found it almost impossible to
kill sql server backups once they are started. From what I can tell, a
single job from the Automatic Backup schedule is started, which then
triggers individual jobs per database using the Default-Policy schedule. I
assumed that killing the 'parent' Automatic Backup job would prevent
further Default-Policy jobs from appearing, but this is not the case. Each
time I kill a Default-Policy, another is queued. As a bodge I can halt the
backups by killing bpinetd on the client, but the jobs start queuing again
when I restart bpinetd. What process should I kill to halt a sql backup
mid-way through?
Cheers,
Paul
"Marianne Van den Berg"
<Marianne.VandenBerg@mgx To: "Paul Sobey"
<Paul.Sobey AT aleagroup DOT com>, <veritas-bu AT mailman.eng.auburn DOT edu>
group.com> cc:
Subject: RE:
[Veritas-bu] NetBackup + MSSQL
25/01/2002 07:05
Hi Paul
Try to increase the "MEDIA_UNMOUNT_DELAY" - see NetBackup DataCenter
System Administrator's Guide - Windows NT/2000 p 518.
Regards
Marianne van den Berg
Senior Support Engineer
MGX Storage Solutions
Tel: +27 11 695 2064
Fax: +27 11 695 2291
Mobile: +27 82 468 8107
mailto: marianne.vandenberg AT mgxgroup DOT com
This e-mail and attachments relating thereto, is intended for the
above-mentioned recipient. If you have received his e-mail in error,
kindly notify the sender and delete it immediately as it contains
information relating to the official business of MGX Holdings Ltd.,
which is confidential, legally privileged and proprietary to MGX
Holdings Ltd. MGX Holdings Ltd does not own and endorse any other
content. The views and opinions expressed in this e-mail are those of
the sender unless clearly stated as being those of MGX Holdings Ltd.
-----Original Message-----
From: Paul Sobey [mailto:Paul.Sobey AT aleagroup DOT com]
Sent: Wednesday, January 23, 2002 11:53 PM
To: veritas-bu AT mailman.eng.auburn DOT edu
Subject: [Veritas-bu] NetBackup + MSSQL
Hi Everyone,
I have just added a MSSQL class to my NetBackup server, and have added
my
first server. The server contains around 127 databases, totalling
approximately 12GB.
While performing a test backup, I noticed that each new database stream
took an average of three minutes before any data was actually written to
tape. When I looked at the server, I discovered that NetBackup was
unloading and reloading the tape into the robot between databases. Is
this
normal? Losing 127 x 3 minutes each time I back up this server amounts
to
an awful lot of wasted time during the backups! I noticed that when I
added
a second SQL server to the class, during the period when both servers
were
backing up concurrently the reloading behaviour didn't occur because
Media
Manager didn't have a chance to unload the tape. I looked for a way of
streaming multiple databases from the same server to achieve the same
effect, but the STRIPES command in the SQL directive batch file seems to
refer to multiple stripes of a single database.
Can anyone offer some advice to a NetBackup newbie? I feel like I'm
missing
something obvious...
Using 3.41 on NT4 with an LTO autoloader.
Cheers,
Paul
_______________________________________________
Veritas-bu maillist - Veritas-bu AT mailman.eng.auburn DOT edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu
|
